Point cloud segmentation is a fundamental problem in point processing. Segmenting a point cloud fully automatically is very chal- lenging due to the property of point cloud as well as different requirements of distinct users. In this paper, an interactive segmentation method for point clouds is proposed. Only two strokes need to be drawn intuitively to indicate the target object and the background respectively. The draw strokes are sparse and don’t necessarily cover the whole object. Given the strokes, a weighted graph is built and the segmentation is formulated as a minimization problem. The problem is solved efficiently by using the Max Flow Min Cut algorithm. In the experiments, the mobile mapping data of a city area is utilized. The resulting segmentations demonstrate the efficiency of the method that can be potentially applied for general point clouds.
Figure 1: The two strokes shown in (a) are drawn by users. The orange one specifies the target object, i.e., a van in the point cloud, and the gray one specifies the background. The strokes are sparse and don’t necessarily cover a large part. Using graph cuts, the selection of the orange stroke can propagate over the whole object and hence the van is segmented out from the point cloud as displayed in (b).
Figure 2: The orange and gray strokes are drawn by users to specify the van and the background. Given the two strokes, a graph of the point cloud is built, the dots represent the nodes in the graph and the dashed lines are the edges. Noticeably, the blue and the orange dots are two added nodes as source and sink.
Figure 3: A 2D example is given to illustrate the algorithm in [Katz et al., 2007] for computing visibility of point cloud. The blue heart represents a point cloud and the view point is in the center of the circle. The blue heart is transformed to the cyan distorted one. A convex hull of the transformed heart and the view point is computed, and the visible part of the blue heart exactly corresponds the hull boundary.